© 2007 The Author(s).
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper is available online free of all access charges (see http://jxb.oxfordjournals.org/open_access.html for further details)
RESEARCH PAPER |
Transpiration efficiency of a tropical pioneer tree (Ficus insipida) in relation to soil fertility
1Smithsonian Tropical Research Institute, PO Box 0843-03092, Balboa, Ancon, Republic of Panama
2Department of Forest Resources, University of Idaho, Moscow, ID 83844-1133, USA
* To whom correspondence should be addressed. E-mail: cernusakl{at}si.edu
Received 6 June 2007; Revised 30 July 2007 Accepted 3 August 2007
 |
Abstract |
|---|
 Top Abstract IntroductionTheoryMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
The response of whole-plant water-use efficiency, termed transpiration efficiency (TE), to variation in soil fertility was assessed in a tropical pioneer tree, Ficus insipida Willd. Measurements of stable isotope ratios (
13C, 18O, 15N), elemental concentrations (C, N, P), plant growth, instantaneous leaf gas exchange, and whole-plant water use were used to analyse the mechanisms controlling TE. Plants were grown individually in 19 l pots with non-limiting soil moisture. Soil fertility was altered by mixing soil with varying proportions of rice husks, and applying a slow release fertilizer. A large variation was observed in leaf photosynthetic rate, mean relative growth rate (RGR), and TE in response to experimental treatments; these traits were well correlated with variation in leaf N concentration. Variation in TE showed a strong dependence on the ratio of intercellular to ambient CO2 mole fractions (ci/ca); both for instantaneous measurements of ci/ca (R2=0.69, P <0.0001, n=30), and integrated estimates based on C isotope discrimination (R2=0.88, P <0.0001, n=30). On the other hand, variations in the leaf-to-air humidity gradient, unproductive water loss, and respiratory C use probably played only minor roles in modulating TE in the face of variable soil fertility. The pronounced variation in TE resulted from a combination of the strong response of ci/ca to leaf N, and inherently high values of ci/ca for this tropical tree species; these two factors conspired to cause a 4-fold variation among treatments in (1–ci/ca), the term that actually modifies TE. Results suggest that variation in plant N status could have important implications for the coupling between C and water exchange in tropical forest trees. Key words: Carbon isotope, oxygen isotope, soil fertility, transpiration efficiency, tropical tree
|
Introduction |
|---|
|
TopAbstractIntroductionTheoryMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Water-use efficiency at the whole-plant level, often referred to as transpiration efficiency (TE), is defined as the rate of biomass production of a plant relative to the rate of transpiration (Bacon, 2004). Although ecophysiologists frequently assess water-use efficiency at the leaf level, relatively few measurements of TE have been reported, largely due to the logistical challenges involved in obtaining such data. Nonetheless, the whole plant is clearly a meaningful organizational level at which to analyse controls over growth, CO2 exchange, and water use (McCree, 1986; Meinzer and Goldstein, 1996). The TE effectively describes the coupling between whole-plant C and water exchange in terrestrial vegetation. Thus, a mechanistic understanding of the controls over TE is relevant to studies of plant competition, ecosystem function, and plant responses to climate change. In tropical forests, it was recently reported that seasonal drought significantly impacts tree community dynamics (Engelbrecht et al., 2007), suggesting that TE could play an important role in determining the performance and distribution of tropical tree species.
Leaf-level water-use efficiency generally increases in response to increasing leaf N concentration in C3 plants (Wong, 1979; Toft et al., 1989; Duursma and Marshall, 2006). This is because more leaf N is usually associated with more photosynthetic capacity, which allows for a greater photosynthetic rate at a given rate of water loss. At the whole-plant scale, the TE of trees has also generally been observed to increase with increasing N availability (Guehl et al., 1995; Syvertsen et al., 1997; Livingston et al., 1999; Ripullone et al., 2004), although not always (Guehl et al., 1995; Hobbie and Colpaert, 2004). This trend also appears to apply for tropical trees: TE of Ficus insipida Willd., a fast-growing tropical pioneer tree capable of high rates of photosynthesis (Zotz et al., 1995), increased in response to fertilizer application (Winter et al., 2001); and a linear relationship was observed between TE and leaf N concentration in an experiment involving seven tropical tree species (Cernusak et al., 2007). Although there appears to be general agreement among experiments regarding the direction of response of tree TE to variation in soil fertility (Raven et al., 2004), the physiological mechanisms modulating the response are less well understood. For example, it was suggested that variation in respiratory C use (Guehl et al., 1995), or in the amount of unproductive water loss (Hobbie and Colpaert, 2004), could play important roles in determining the response of TE to soil fertility, in addition to the effects associated with leaf-level photosynthesis.
Since the revelation that C isotope fractionation correlates positively with the ratio of intercellular to ambient CO2 mole fractions in C3 plant leaves (Farquhar et al., 1982), analysis of 13C/12C in plant organic material has played an important role in water-use efficiency research (Bacon, 2004). It was later suggested that analysis of 18O/16O in plant organic material could also aid investigations of water-use efficiency by providing complementary information to that obtained from C isotope analyses (Farquhar et al., 1989; Sternberg et al., 1989). In this study, measurements of C and O isotope ratios were combined with measurements of plant elemental composition, growth, instantaneous leaf gas exchange, and whole-plant water use to analyse the mechanisms controlling the response of TE of the tropical pioneer tree Ficus insipida to variation in soil fertility.
|
Theory |
|---|
|
TopAbstractIntroductionTheoryMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
At the leaf level, photosynthetic water-use efficiency can be expressed as the quotient of the diffusive fluxes of CO2 and water vapour into and out of the leaf, respectively, during photosynthesis (Farquhar and Richards, 1984):
 | (1) |
Equation (1) can be scaled to the whole-plant level by taking into account respiratory C use and water loss not associated with photosynthesis (Farquhar and Richards, 1984; Hubick and Farquhar, 1989). Thus, whole-plant transpiration efficiency (TE) can be defined as
 | (2) |
c is the proportion of C fixed during photosynthesis that is subsequently lost by respiration from roots and stems during the day, and from roots, stems, and leaves during the night; and w is the proportion of unproductive water loss relative to that associated with C uptake, i.e. nocturnal transpiration through partially open stomata and cuticular water loss by leaves and stems during the day and night.
|
The ratio of intercellular to ambient CO2 mole fractions (ci/ca), shown in
equation (2), also relates independently to C isotope discrimination (
13C). The 13C for C3 photosynthesis can be defined as (Farquhar et al., 1982; Farquhar and Richards, 1984; Hubick et al., 1986)
 | (3) |
), and b is the fractionation caused by Rubisco, the primary carboxylating enzyme in C3 plants (29). The term d summarizes collectively the fractionations caused by dissolution of CO2 and liquid phase diffusion, photorespiration, and dark respiration (Farquhar et al., 1989). The effects of fractionations associated with these processes on the overall 13C are small compared with that caused by Rubisco, but nonetheless significant (Brugnoli and Farquhar, 2000; Ghashghaie et al., 2003). The d in equation (3) substitutes for the following terms in the full model of 13C for C3 plants (Farquhar et al., 1989)
 | (4) |
), es is that during dissolution into water (1.1), and al is that during liquid phase diffusion (0.7). The Rd is day respiration (µmol CO2 m–2 s–1), e is the 13C discrimination associated with day respiration, k is the carboxylation efficiency (mol CO2 m–2 s–1), f is the discrimination against 13C associated with photorespiration, and 
* is the CO2 compensation point in the absence of Rd (µmol mol–1). The 13C in equation (3) is defined with respect to atmospheric CO2 as 13C=Ra/Rp–1, where Ra is 13C/12C of atmospheric CO2 and Rp is 13C/12C of plant material (Farquhar and Richards, 1984). In practice, 13C is calculated from measured 13C values as
 | (5) |
13Ca is 13C of CO2 in air, and 13Cp is that of plant material. For convenience, 13C and 13C values are typically expressed as per mil (), meaning that they have been multiplied by the scaling factor 1000.
Equations (2) and (3) suggest that TE and 13C share a mutual dependence on ci/ca. Combining the two equations yields (Hubick and Farquhar, 1989)
 | (6) |
 | (7) |
Equation (7) presents a linear relationship between TE and 13C with slope –m and intercept m(b–d):
 | (8) |
c)/[1.6v(1+w)(b–a)]. Thus, as demonstrated previously (Hubick et al., 1986), the coefficients of a linear regression equation between TE and 13C can be used to make inferences about parameters in equation (7) that are difficult to determine experimentally. Namely, the term d can be calculated as d=b–(I/m), where I is the intercept and m is the negative slope of the relationship between TE and 13C, and c can be calculated as c=1–1.6vm(1+w)(b–a)/ca. Note that these calculations assume that the terms for which m and I substitute in equation (7) are invariant over the range of 13C for which the regression equation is fitted. It was previously suggested that measurements of 18O/16O of plant organic material could prove useful in water-use efficiency studies by providing a means for making integrated estimates of v (Farquhar et al., 1989; Sternberg et al., 1989). The v is defined as wi–wa, where wi is the water vapour mole fraction in the leaf intercellular air spaces and wa is that in the surrounding atmosphere. The suggestion was that the following set of equations, describing the processes contributing to the 18O/16O of plant organic material, could be inverted to solve for wi.
The terms wa and wi relate to steady-state leaf water 18O enrichment at the sites of evaporation in leaves (18Oe) in the following way (Craig and Gordon, 1965; Dongmann et al., 1974; Farquhar and Lloyd, 1993)
 | (9) |
+ is the equilibrium fractionation that occurs during the phase change from liquid water to vapour, k is the kinetic fractionation that occurs during water vapour diffusion through stomatal pores and the leaf boundary layer, and 18Ov is the 18O enrichment of atmospheric water vapour with respect to water taken up by the roots (source water). The 18O enrichment (18O) is defined with respect to source water as 18O=R/Rs–1, where R is 18O/16O of the sample of interest and Rs is that of source water. The equilibrium fractionation, +, can be calculated as follows (Bottinga and Craig, 1969):
 | (10) |
k, can be calculated as (Farquhar et al., 1989b)
 | (11) |
The 18O of leaf mesophyll water (18OL), the signature most relevant to production of plant organic material (Cernusak et al., 2003), can be related to 18Oe as (Farquhar and Lloyd, 1993; Farquhar and Gan, 2003)
 | (12) |
is a Péclet number, defined as EL/(CD), where E is transpiration rate (mol m–2 s–1), L is a scaled effective path length (m), C is the molar concentration of water (mol m–3), and D is the diffusivity of H218O in water (m2 s–1). The D can be calculated as (Cuntz et al., 2007)
 | (13) |
18OL can in turn be related to the 18O enrichment of plant cellulose (18Oc) according to the following equation (Barbour and Farquhar, 2000):
 | (14) |
wc is the equilibrium fractionation between organic O and medium water. Finally, the 18O enrichment of plant dry matter (18Op) can be related to that of plant cellulose by adding an additional fractionation factor (cp) to account for the 18O difference between the two (Barbour and Farquhar, 2000);
 | (15) |
|
Materials and methods |
|---|
|
TopAbstractIntroductionTheoryMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Plant material and experimental treatments
The experiment took place at the Smithsonian Tropical Research Institute, Santa Cruz Experimental Field Facility, Gamboa, Republic of Panama. The site is located at 9°07’ N latitude, 79°42’ W longitude, at an altitude of 28 m above sea level. Seeds of Ficus insipida Willd. (Moraceae) were collected from mature trees growing in the Panama Canal watershed. Seeds were germinated in trays containing a commercial potting soil in July 2005. Following germination, three seedlings each were transplanted into 40 pots, each of volume 19 l. Upon transplanting, a handful of soil was added to each pot together with roots taken from the base of the palm tree Attalea butyracea (Mutis ex L.f.) Wess. Boer. as a source of mycorrhizal inoculant. The pots were placed under a translucent rain shelter on plastic tables; they were elevated approximately 0.8 m above the concrete surface below the shelter. The shelter reduced incoming photon flux density (PFD) by approximately 20%. After an adjustment period of about 1 month, two seedlings were removed from each pot, leaving a uniform population of seedlings, with one seedling per pot. Among the 40 pots, five treatments were deployed, yielding eight pots per treatment. For each treatment, two pots were selected to serve as controls without plants; seedlings were removed from these pots. The discarded seedlings were used to measure seedling dry mass at the beginning of the experiment, estimated as 0.27 g.
The five soil fertility treatments consisted of varying mixtures of homogenous, dark topsoil and rice husks, with the high fertility treatment additionally receiving a one-time application of slow-release fertilizer. It was expected that the addition of rice husks to the soil mixture would reduce the soil fertility in two ways; both by diluting the nutrient content of the pot, and by adding a high C/N substrate that would tend to immobilize N and other nutrients, leading to greater deficiencies as the proportion of rice husks increased. The treatments were as follows, given as the volumetric percentage of air-dried topsoil in the topsoil/rice-husk mixture: 20, 40, 60, 80, 80+N. For the 80+N treatment, approximately 13 g of Osmocote-Plus controlled-release fertilizer (Scotts-Sierra, Maryville, OH, USA) was added to each pot. Due to the difference in density between the topsoil and rice husks, the dry mass of the topsoil/rice-husk mixture required to fill each 19 l pot varied by treatment: 5.4, 8.9, 12.4, 15.4, and 15.4 kg were placed in each pot for treatments 20, 40, 60, 80, and 80+N, respectively. The amount of water required to bring the pots to field capacity also varied slightly: 4.0, 4.5, 5.0, 5.0, and 5.0 kg of water were added to treatments 20, 40, 60, 80, and 80+N, respectively. 1.5 kg of gravel was added to the soil surface of each pot to reduce soil evaporation.
Plant water use
The pots were weighed at regular intervals from 23 August 2005 until plant harvest on 4 November 2005, a period of approximately 10 weeks. Pot weights were determined with a 64 kg capacity balance (Sartorius QS64B, Thomas, Swedesboro, NJ, USA). The pots were initially weighed once per week, but the frequency was increased to as much as three times per week when plant stature and water use increased toward the end of the experiment. Pot water loss in the interval between weight measurements did not exceed 2.5 kg; this value was only approached near the very end of the experiment and only for the largest plants. Woody tree seedlings typically do not show a reduction in daily transpiration rate until soil water content falls to approximately one-third the value at field capacity (Sinclair et al., 2005), so it is assumed that transpiration was not limited by soil water content at any time during the experiment. After weighing each pot, water was added until the initial weight at field capacity was restored. Plant transpiration over the course of the experiment was calculated as the difference between cumulative pot water loss and the mean water loss of the control pots for each treatment. Leaves, stems, and roots were oven-dried at 70 °C after harvest and weighed separately for each plant. Abscised leaves were collected during the experiment and their dry weight added to the plant dry weight for TE calculations. Leaf area at plant harvest was determined with an LI-3100 Leaf-Area Meter (Li-Cor Inc., Lincoln, NE, USA).
Meteorological conditions during the experiment were recorded every 15 min using an automated weather station (Campbell Scientific, Logan, UT, USA), as described previously (Winter et al., 2001, 2005). The mean daytime temperature, calculated between the h of sunrise and sunset, was 27.3 °C; mean daytime relative humidity was 81.5%; mean PFD was 670 µmol m–2 s–1; and mean daytime wind speed was 0.4 m s–1.
For three days prior to plant harvest, daily and nightly water use of each plant was measured. Pots were weighed prior to sunrise (05.30 h) and again following sunset (18.00 h).
Control pots were also weighed and control water loss subtracted from that of pots with plants to calculate plant water loss. Mean daily and nightly transpiration rates were expressed on a leaf area basis by dividing by the leaf area determined at plant harvest, which followed the third cycle of day/night measurements. The term w for each plant was calculated as night-time plant water use divided by daytime plant water use.
Leaf gas exchange measurements
Gas exchange of the youngest, fully-expanded leaf of each plant was measured under light-saturating conditions (PFD >800 µmol m–2 s–1) at both morning and midday on 20 October 2005 with an Li-6400 portable photosynthesis system (Li-Cor Inc., Lincoln, NE, USA). Leaves were illuminated during measurements by natural sunlight. The mean PFD at the leaf surface during morning measurements was 1094±225 µmol m–2 s–1 (mean ±1 SD), whereas that during midday measurements was 1333±121 µmol m–2 s–1. The mean v during morning measurements was 12.8±1.6 mmol mol–1, and that during midday measurements was 17.2±1.4 mmol mol–1. Mean leaf temperatures (TL) during morning and midday measurements were 33.0±1.0 °C, and 36.0±0.5 °C, respectively. Dark respiration was measured on the youngest, fully-expanded leaf of each plant on 3 November 2005 between 19.30 h and 22.00 h. Mean leaf temperature during measurements was 25.9±0.2 °C.
Isotopic and elemental analyses
Leaf, stem, and root dry matter were ground to a fine powder for elemental and isotopic analyses. The 13C/12C and 15N/14N isotope ratios of leaf, stem, and root dry matter were measured at the Idaho Stable Isotopes Laboratory at the University of Idaho, Moscow, ID, USA; the 18O/16O of leaf dry matter was measured at the Stable Isotope Core Laboratory, Washington State University, Pullman, WA, USA. For 13C/12C and 15N/14N analyses, samples of approximately 3 mg were combusted in an NC2500 elemental analyser (CE Instruments, Milan, Italy), then swept by a helium carrier gas, via a continuous flow interface, into a Delta Plus isotope ratio mass spectrometer (Finnigan MAT, Bremen, Germany). In addition to 13C/12C and 15N/14N, the C and N elemental concentrations of the sample material were determined from peak areas obtained from mass spectrometric measurements. The 18O/16O of leaf dry matter was measured on samples of approximately 1 mg on a Delta XP isotope ratio mass spectrometer (Finnigan MAT, Bremen, Germany), following pyrolysis in a high-temperature furnace (Thermoquest TC/EA, Finnigan MAT, Bremen, Germany). The C, N, and O stable isotope ratios were obtained in delta notation relative to standards of Pee Dee Belemnite, air, and Vienna Standard Mean Ocean Water, respectively.
Whole-plant 13C and 15N compositions were calculated by mass balance using the dry mass of each plant organ (leaves, stems, and roots), the C or N mass fraction, and the 13C or 15N composition. The 13C was calculated from 13C values according to equation (5). The 13Ca was assumed to be –8, consistent with observed daytime 13Ca near Panama City, Panama (Winter and Holtum, 2002). The leaf 18Op was expressed with respect to the 18O of irrigation water (–4.0) according to the equation 18Op=(18Op–18Os)/(1+18Os), where 18Op is the 18O of leaf dry matter, and 18Os is that of irrigation (source) water (Barbour et al., 2004).
In addition, elemental concentrations of P, K, Ca, Mg, Mn, and Zn were quantified in leaf dry matter. Approximately 200 mg of finely ground, oven-dried leaf material were digested at 380 °C in sulphuric acid and lithium sulphate with a selenium catalyst and hydrogen peroxide. Elemental concentrations were then determined on an inductively-coupled plasma optical-emission spectrometer (Perkin Elmer Inc., Wellesley, MA, USA).
Leaf temperature and leaf-to-air humidity gradient
Three different methods were used to estimate average values for TL and v over the course of the experiment for the five soil fertility treatments. In the first method, a leaf energy balance model was used, details of which have been recently described (Barbour et al., 2000; Cernusak et al., 2003a). In the model, mean daytime air temperature, relative humidity, irradiance, and wind speed were used over the course of the experiment along with the transpiration rates measured gravimetrically just prior to plant harvest. The transpiration rates used in the analysis were those measured on 2 November 2005, when the mean daily PFD and mean daily air water vapour mole fraction deficit matched very closely the average values recorded over the course of the experiment (684 µmol m–2 s–1 and 6.6 mmol mol–1 versus 670 µmol m–2 s–1 and 6.6 mmol mol–1, respectively). It was assumed that the incoming PFD was reduced by 20% by the translucent rain shelter covering the plants, and that the mean intercepted PFD for each plant was further reduced by 25% by self shading and non-horizontal leaf orientation. Thus, the PFD used in the leaf energy balance analysis was 400 µmol m–2 s–1. The mean surface area of individual leaves for each treatment was used to calculate boundary layer conductance. The leaf energy balance model was used to predict mean TL for each treatment, and v was calculated as the difference between the saturation vapour mole fraction at TL and the average daytime air vapour mole fraction.
In the second method for estimating TL and v, equation (2) was used, along with measured values of TE, ci/ca (based on measurements of 13C), and w. It was assumed that ca was constant at 375 µmol mol–1, and that c was constant at 0.4. Equation (2) for v was then solved. Average wa was used to calculated wi, and TL was calculated as the dew point temperature at wi.
The third method for estimating TL and v was based on measurements of 18Op for leaf dry matter. Equations (9) to (15) were inverted to solve for wi starting with values of 18Op. The cp was assumed to be –6.8 (Cernusak et al., 2004); wc was assumed to be 27 (Sternberg and DeNiro, 1983); the term pexpx was assumed to be 0.38 (Cernusak et al., 2005); the L was assumed to be 0.015 m (Cernusak et al., 2002, 2005); the D was calculated to be 2.46x10–9 m2 s–1; the rs values were taken from leaf gas exchange measurements; the rb was calculated from mean leaf surface area and mean wind speed, as for the leaf energy balance analysis; the + was calculated to be 8.9; and the 18Ov was assumed equal to –+ (Farquhar et al., 2007). Although the parameters D and + are dependent on leaf temperature, estimates of v are relatively insensitive to small changes in these parameters. For example, calculating these parameters with a TL of 25 °C versus 30 °C would shift our mean estimate of v from 8.1 to 8.0 mmol mol–1 in the case of D, and from 7.9 to 8.0 mmol mol–1 in the case of +. Therefore, D and + were calculated assuming an approximate TL of 28 °C. Final TL was calculated from wi as described above.
Statistical analysis
Ordinary least squares regression was used to evaluate the relationships between leaf isotopic characteristics, gas exchange characteristics, and leaf elemental concentrations. These analyses were used to determine the ability of an explanatory variable to predict variation in a response variable. However, to determine the regression coefficients for the relationship between TE and 13C for estimations of d and c, geometric mean regression was used. In this case, we were interested in the functional relationship between TE and 13C rather than a causal, predictive relationship, the values of the regression coefficients were the primary focus of the analysis, and both parameters were measured with error, thereby indicating the use of geometric mean regression (Sokal and Rohlf, 1995). Analysis of variance was used to test for variation among treatments in physiological and morphological parameters. Tukey's method for pair-wise comparisons was used to test for significant differences between individual treatments.
|
Results |
|---|
|
TopAbstractIntroductionTheoryMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Plant growth and morphology
Plant growth and morphology differed among the soil fertility treatments (Table 2). Mean plant dry mass at the time of harvest ranged from 2.6 to 80.4 g from the lowest to the highest soil fertility treatment. Corresponding mean relative growth rates (RGR) ranged from 30.4 to 78.0 mg g–1 d–1, respectively (Table 2). Differences in RGR among treatments were statistically significant for all pairs of treatments, except 40% and 60% soil. There was significant variation in root:shoot ratio, leaf area ratio, and specific leaf area among treatments, but this variation did not appear to be systematically related to the soil fertility treatments (Table 2).
|
Elemental composition
Whole-plant C concentration varied among treatments, but over a rather narrow range of 0.40–0.42 g g–1 (Table 3). Whole-plant N concentration, on the other hand, showed more pronounced variation among treatments, ranging from 11.6 to 19.7 mg g–1 (Table 3). Leaf N concentration per unit leaf area (Narea) increased from 53.9 mmol m–2 to 90.4 mmol m–2 from the lowest to the highest soil fertility treatment (Table 3). Leaf P concentration showed an opposite trend to leaf N concentration, decreasing from the lowest to the highest soil fertility treatment (Table 3). Leaf Mg, Mn, and Zn concentrations followed similar trends to leaf P, decreasing from the lowest to the highest soil fertility (Table 3). Leaf K and Ca concentrations were less variable among treatments, although they showed weak tendencies to decrease (K) or increase (Ca) from the lowest to the highest soil fertility (Table 3).
|
The mass ratio of leaf N to P (N/P) increased across treatments from a mean of 4.7 for the lowest soil fertility to 16.4 for the highest soil fertility (Table 3). Variation in TE was closely correlated with variation in N/P (Fig. 1A). The relationship between the two parameters was slightly non-linear, such that the natural logarithm of N/P explained 90% of variation in TE, whereas N/P explained 87%. The relationship between Ln(N/P) and TE was TE=1.42Ln(N/P)–1.28 (R2=0.90, P <0.0001, n=30). The RGR showed a non-linear response to variation in N/P; it increased up to N/P of about 15, then decreased slightly over the small range of values above 15 (Fig. 1B). Variation in RGR and TE was closely correlated (Fig. 1C). Again the relationship was slightly non-linear, such that the relationship between RGR and Ln(TE) was slightly stronger (R2=0.95, P <0.0001, n=30) than that between RGR and TE (R2=0.92, P <0.0001, n=30).
|
Leaf Parea showed a strong positive, linear dependence on mean daytime transpiration rate, also expressed on a leaf area basis, with Egrav explaining 74% of variation in leaf Parea (Fig. 2A). Similar positive, linear dependencies on Egrav were also observed for leaf Zn per unit area (R2=0.45, P <0.0001, n=29) and leaf Mg per unit area (R2=0.23, P <0.005, n=29).
|
Leaf gas exchange
Photosynthesis of the youngest, fully-expanded leaf of each plant at saturating irradiance increased with increasing soil fertility; mean treatment values ranged from 11.0 to 22.4 µmol CO2 m–2 s–1 from the lowest to the highest soil fertility treatment (Table 2). These values represent averages of measurements made during morning and midday. Leaf dark respiration also increased from low to high soil fertility, with treatment means ranging from 0.71 to 1.34 µmol CO2 m–2 s–1 (Table 2). The ratio of leaf dark respiration to leaf net photosynthesis, on the other hand, was invariant among treatments (Table 2). Leaf net photosynthesis, leaf dark respiration, and whole-plant relative growth rate were linearly related to leaf Narea (Fig. 3). The ratio of leaf dark respiration to leaf net photosynthesis, in contrast, showed no correlation with Narea (P=0.67, n=30).
|
1200 µmol m–2 s–1. Leaf dark respiration was measured on 3 November 2005; mean leaf temperature was Gravimetric transpiration
Mean daytime transpiration rates, determined gravimetrically, varied among soil fertility treatments. Mean treatment daytime Egrav ranged from 1.92 to 1.14 mmol m–2 s–1, and generally decreased from the lowest soil fertility treatment to the highest (Table 2). Mean night-time Egrav followed a similar pattern (Table 2). The term
w, describing night-time transpiration as a proportion of daytime transpiration, ranged from 0.15 to 0.11 from low to high soil fertility treatments, respectively (Table 2). Mean daytime Egrav showed a negative, linear dependence on Narea. The equation describing the relationship was Egrav= –0.01Narea+2.3 (R2=0.29, P=0.001, n=29).
Stable isotope composition
Mean values of 13CL, 13CS, 13CR, and 13Cwp for the five soil fertility treatments are shown in Table 2. All 13C values decreased from lowest to highest soil fertility (Table 2). The general pattern among the different plant tissues was 13CL>13CS>13CR (Table 2). Average whole-plant 15N values spanned a relatively narrow range from 2.4 to 3.4, and did not differ significantly among treatments (Table 2). Leaf 18Op was significantly lower for the highest soil fertility treatment than for the other treatments, and tended to decrease with increasing soil fertility (Table 2). The 18Op was negatively correlated with mean daytime Egrav across all treatments (Fig. 2B).
Leaf temperature and leaf-to-air humidity gradient
Estimates of TL and v generated by the three different methods are summarized in Table 4. The leaf energy balance model estimated a variation of 1.5 °C in TL across the soil fertility treatments. Values ranged from 27.1 °C to 28.6 °C from the lowest to the highest soil fertility treatment; corresponding values for v ranged from 6.2 mmol mol–1 to 9.5 mmol mol–1 (Table 4). The second method that was used to estimate TL and v, which assumed a constant value of 0.4 for c, predicted less variation among treatments; values ranged from 26.9 °C to 27.8 °C and 5.8 mmol mol–1 to 7.8 mmol mol–1, respectively. The third method, based on measurements of 18Op, predicted a similar range of values to the leaf energy balance model, but with values trending in the opposite direction across treatments; i.e., decreasing from the lowest to the highest soil fertility. For the 18Op method, values ranged from 28.4 °C to 27.0 °C for TL and from 9.0 mmol mol–1 to 6.0 mmol mol–1 for v (Table 4).
|
Transpiration efficiency
The TE varied strongly among soil fertility treatments (Table 2). Variation in TE was closely correlated with variation in
13Cwp (Fig. 4). The equation relating TE to 13Cwp, determined by geometric mean regression, was TE= –0.7213Cwp+18.0. These coefficients were similar to those determined by ordinary least-squares regression (–0.68 and 17.0 for m and I, respectively). The d calculated from m and I values of –0.72 and 18.0 was 4.0. Long-term, integrated ci/ca estimated from 13Cwp, using d=4.0, varied strongly among treatments (Table 2). Treatment means ranged from 0.96 to 0.86 from low to high soil fertility. The 13Cwp was well correlated with ci/ca determined from instantaneous gas exchange measurements (Fig. 5), as was TE (R2=0.69, P <0.0001, n=30). Instantaneous ci/ca was lower than 13Cwp-based ci/ca for all treatments (Table 2). Instantaneous ci/ca, 13Cwp, and TE each correlated well with leaf Narea (Fig. 6A, B, C). These parameters also correlated strongly with the light-saturated photosynthetic rate of the youngest, fully-expanded leaf of each plant (Fig. 6D, E, F). There were weaker correlations between stomatal conductance and ci/ca, 13Cwp, and TE; however, these correlations were opposite in sign to what would be expected if variation in stomatal conductance were controlling variation in these parameters (Table 2).
|
|
|
The transpiration efficiency of N acquisition (TEN), calculated as whole-plant N increment divided by cumulative plant water loss, increased from 17.8 µmol N mol–1 H2O for the lowest soil fertility treatment to 107.0 µmol N mol–1 H2O for the highest soil fertility treatment (Table 2). The TE increased linearly as a function of TEN according to the equation TE=0.020TEN+0.58 (R2=0.84, P <0.0001, n=30).
A sensitivity analysis is shown in Table 5 illustrating the predicted effect of changing the input parameters in equation (2) on TE. The ranges of input values used in the analysis were based on observations for the five soil fertility treatments for the parameters ci/ca, v, and w; for the parameters c and ca, the selected ranges represent best guesses at likely limits of their variation. Table 5 suggests that the variation that was observed in TE was largely driven by variation in ci/ca.
|
|
Discussion |
|---|
|
TopAbstractIntroductionTheoryMaterials and methodsResultsDiscussionConclusionsReferences |
|---|
Large variation in growth and whole-plant water-use efficiency of a tropical pioneer tree in response to variation in soil fertility was observed under non-limiting soil moisture conditions. Analyses of elemental concentrations in dry matter of the experimental plants indicated that treatment differences in RGR and TE resulted largely from variation in N availability. Leaf photosynthesis and dark respiration rates were well correlated with leaf Narea, as was RGR (Fig. 3). Variation in TE was linearly correlated with variation in ci/ca, both for instantaneous measurements of ci/ca, and for integrated estimates based on
13Cwp (Figs 4, 5). The ci/ca, 13Cwp, and TE, in turn, were well correlated with variation in leaf Narea and leaf photosynthesis (Fig. 6). The response of TE to soil fertility was largely caused by variation in ci/ca; on the other hand, variations in v, c and w probably played only minor roles in modulating TE in response to soil fertility (Table 5). The variation in ci/ca resulted from variation in photosynthetic capacity caused by variation in leaf Narea, rather than from variation in stomatal conductance (Table 2; Fig. 6).
Elemental composition
Of the mineral elements quantified in the leaves of the experimental plants, leaf N showed the strongest positive relationship with RGR. The only other element to be positively correlated with RGR was Ca. However, leaf Ca per unit area explained only 13% of variation in RGR, whereas leaf N per unit area explained 46%. Thus, it is concluded that plant growth was primarily constrained by N availability. Of the other measured elements, P, Mg, and Zn showed positive linear correlations with Egrav, suggesting that these elements were absorbed in relatively constant proportion to the water flux into the roots. The relationship with leaf P was particularly striking, with variation in Egrav explaining 74% of variation in leaf Parea (Fig. 2A). This relationship was surprising, given that P is generally thought to be relatively insoluble in soils. Supply of a source of mycorrhizal inoculant at planting and the high organic matter content of the soil due to the addition of rice husks may have played some role in enabling the relationship between leaf Parea and Egrav in our experiment. In contrast to the relationship between leaf Parea and Egrav, there was a negative correlation between leaf Narea and Egrav across the soil fertility treatments. Variation in leaf Narea explained 29% of variation in Egrav. Increases in transpiration in response to low N availability have also been observed in other tree species (Guehl et al., 1995; Livingston et al., 1999), and transpirational control of N accumulation was previously demonstrated in a mistletoe/tree complex (Marshall et al., 1994).
A non-linear response of RGR to leaf N/P was observed (Fig. 1B). The maximum RGR occurred near N/P of 15, which agrees well with the prediction that N and P should be in balanced supply at N/P between 14 and 16 (Koerselman and Meuleman, 1996). Koerselman and Meuleman (1996) suggested that at N/P <14, growth would be N limited, whereas at N/P >16, growth would be P limited. This further reinforces the notion that variation in RGR and TE in our experiment was primarily caused by differential N availability.
It was found that N/P was an excellent predictor of variation in TE (Fig. 1A); the natural logarithm of N/P explained 90% of variation in TE. For comparison, 13Cwp explained 88% of variation in TE. The close correlation between N/P and TE stemmed from the relationships between leaf N and photosynthetic rate (Fig. 3A) and leaf P and transpiration rate (Fig. 2A). Whether such a relationship will hold up more generally outside our experimental conditions is unknown.
C isotope discrimination and ci/ca
Estimating ci/ca from 13C according to equation (3) requires an estimate of d. The method described in the theory section for estimating d, based on the coefficients of a regression analysis between TE to 13C, assumes that the terms for which the slope and intercept coefficients substitute are invariant over the range of the analysis. This assumption was not strictly met in our experiment; for example, there were probably subtle variations in v and w among treatments (Tables 2, 4). However, if the assumption were strongly violated, one would expect either to see curvature in the relationship between TE and 13Cwp, or a large degree of scatter. In fact, the relationship that was observed appeared linear with little scatter: variation in 13Cwp explained 88% of variation in TE in a least-squares linear regression (Fig. 4). There were, however, two data points that appeared to depart slightly from the linear trend; these two points had the highest 13Cwp values in the dataset (Fig. 4). Repeating the analysis with these two data points excluded would result in an estimate for d of 4.4, whereas the estimate for the full data set was 4.0. Although this difference is not large, it nonetheless highlights the sensitivity of our method for estimating d to variations in the values of the regression coefficients.
Few direct estimates of d exist in the literature, although assessment of d is implicit in the determination of mesophyll conductance from instantaneous measurements of 13C and ci/ca (Evans et al., 1986). Hubick et al. (1986) estimated d to be approximately 3, based on the simultaneous measurements in wheat of 13C and ci/ca (Evans et al., 1986). The d was later estimated to be near zero for barley (Hubick and Farquhar, 1989), and approximately 1 for peanut (Hubick, 1990). Thus, our estimate of 4.0 for d in Ficus insipida is slightly higher than values previously reported for crop plants. In the method used here to calculate d [i.e., d=b–I/m from equation (8)], the value of d clearly depends on the assumed value of b. Fortunately, any change in the assumed value of b has only a minor effect on 13C-based estimates of ci/ca. This is because any change in the assumed value of b will be offset by a similar change in the calculated value of d. Thus, changing the assumed value of b from 29 to 27 in our analysis would only change the mean 13C-based estimate of ci/ca from 0.913 to 0.906.
It is clear from equation (4) that d is a complex parameter, and the general trend in the literature has been to drop it from equation (3) when making long-term, integrated estimates of ci/ca from measurements of 13C in plant tissues. However, based on our analysis, it is suggested that it may not be prudent to omit d from equation (3). Assuming b=29, excluding d from the calculations would shift the mean 13Cwp-based estimate of ci/ca in our experiment from 0.91 to 0.75. Whereas if we assumed b=27, the mean -based estimate of ci/ca would shift from 0.91 to 0.82 if d were omitted. If these ci/ca estimates were then used to predict TE from equation (2), the shift caused by omitting d would equate to an approximately 3-fold increase in predicted TE at b=29, and a doubling of predicted TE at b=27.
It was observed that instantaneous measurements of ci/ca were consistently lower than 13Cwp-based estimates across treatments (Table 2). The mean instantaneous estimate for all treatments combined was 0.85, whereas the mean 13Cwp-based estimate was 0.91. This discrepancy may have partly resulted from differences between v and PFD averaged over the course of the experiment, as compared with values in the cuvette during instantaneous measurements (8 mmol mol–1 versus 15 mmol mol–1 for v, respectively; and 400 µmol m–2 s–1 versus 1200 µmol m–2 s–1 for PFD, respectively). However, at least part of the discrepancy between instantaneous ci/ca and 13Cwp-based ci/ca relates to the fact that ci/ca is calculated differently from instantaneous gas exchange measurements than from isotopic measurements, as described by GD Farquhar (unpublished presentation, BASIN meeting, Marshall, USA, 2004). The equations used to calculate ci from instantaneous measurements are based on a ternary system of gases: CO2, water vapour, and air (Jarman, 1974). The ci is calculated as ci=[(gc–E/2)ca–A]/(gc+E/2), where gc is the total conductance to CO2 of stomata plus boundary layer (Caemmerer and Farquhar, 1981). This calculation takes into account not only collisions between CO2 and air, but also collisions between CO2 and water vapour. By contrast, the equations presented in the theory section of this paper describing the relationship between ci/ca and 13C do not take into account collisions between CO2 and water vapour. Instead, ci is simply defined as ci=ca–A/gc. If the data from the instantaneous measurements are recalculated using this latter definition of ci, a mean value for ci/ca of 0.88 is obtained, which cuts in half the observed difference between instantaneous and 13Cwp-based estimates of ci/ca.
In our analysis of the relationship between TE and 13C, the whole-plant 13C was used rather than that of a single tissue, such as leaves. Using 13CL in place of 13Cwp has only a small effect on our results. It would shift our estimate of d from 4.0 to 3.8, and would shift the mean 13C-based estimate of ci/ca from 0.91 to 0.92.
Unproductive water loss
The unproductive water loss described by the term w in equation (2) comprises non-stomatal water loss during the day and night, and stomatal water loss at night. Of the two, the expectation was that transpiration at night through partially open stomata would dominate. Conductances of leaf cuticles are on the order of 1–2 mmol m–2 s–1 (Kerstiens, 1996), as are surface conductances of branches and stems (Cernusak and Marshall, 2000; Cernusak et al., 2001). In contrast, it is not uncommon to observe nocturnal stomatal conductances ranging from 20 to more than 100 mmol m–2 s–1 (Donovan et al., 1999; Snyder et al., 2003; Bucci et al., 2004; Barbour et al., 2005; Seibt et al., 2007). The mean nocturnal stomatal conductance that was observed during the dark respiration measurements was 124±43 mmol m–2 s–1 (mean ±1 SD). However, because nocturnal v is typically low (mean value of 1.3 mmol mol–1 in our experiment), as are nocturnal wind speeds, nocturnal transpiration is